Jensen Huang, CEO of semiconductor giant Nvidia—whose value has surged 300% over two years—personifies the AI mania. Despite the hype, Huang sought to calm fears of an AI bubble during a recent earnings call, asserting Nvidia sees something very different. This sentiment echoes through other prominent figures: White House AI czar and venture capitalist David Sacks calls the AI surge an investment super-cycle; Silicon Valley investor Ben Horowitz dismisses bubble concerns based on demand and supply fundamentals; and JPMorgan executive Mary Callahan Erdoes deems the bubble idea “crazy, ” highlighting a major operational revolution underway. However, Paul Kedrosky, MIT research fellow and venture capitalist, questions the sustainability behind the frenzy, noting that while AI technology is useful, its rapid improvement has stalled, making expectations of continuous revolution over the next five years misguided. The influx of capital is staggering. OpenAI’s CEO Sam Altman claims $20 billion in annual revenue and plans $1. 4 trillion in data center investments over eight years, contingent on expanding customer adoption. Yet studies reveal limited chatbot impact on firms’ profits and only 3% of users pay for AI services. Nobel laureate economist Daron Acemoglu warns of exaggerations despite potential future productivity gains. Meanwhile, Amazon, Google, Meta, and Microsoft will reportedly spend about $400 billion this year on AI infrastructure, dedicating up to half their cash flow to building data centers. Such spending equates to over $250 from every iPhone user worldwide, clearly unrealistic. To preserve cash, companies like Meta and Oracle are leaning on private equity and debt to finance expansions. Goldman Sachs analysts report a 300% rise in debt by hyperscalers, with recent creative financing via special purpose vehicles (SPVs) used to keep debt off big tech balance sheets. For instance, Wall Street’s Blue Owl Capital and Meta funded a $27 billion Louisiana data center through an SPV, where Meta enjoys full computing capacity but only owns 20% and hides the debt from its books. If the AI bubble bursts, Meta would owe billions regardless of usage.

Investor Gil Luria cautions that while SPVs are now transparent, relying heavily on such structures recalls past collapses like Enron’s and could threaten future stability. This massive debt accumulation assumes future AI revenues will cover costs, yet Morgan Stanley projects Big Tech will spend $3 trillion on AI infrastructure by 2028, with only half funded by operating cash flow. Luria warns that overbuilt capacity coupled with stagnant market growth could render debt worthless, potentially triggering a new financial crisis akin to the dot-com bubble, which similarly collapsed after overinvestment in fiber-optic infrastructure. Further concerns stem from circular investment deals inflating AI demand artificially. Nvidia’s recent $100 billion deal with OpenAI involves Nvidia funding OpenAI’s data centers, which buy Nvidia chips, blurring genuine market signals. Kedrosky notes such arrangements are unusual at this scale and reminiscent of dot-com era hype. Smaller firms like CoreWeave have also entered this loop, exchanging chip capacity and stock with OpenAI, while Nvidia, a CoreWeave investor, agrees to purchase any unused capacity through 2032. Economist Acemoglu warns these deals risk becoming a fragile “house of cards. ” Notably, some top investors are growing wary: Peter Thiel recently sold his entire Nvidia holdings, and SoftBank offloaded nearly $6 billion worth. Michael Burry, famed for predicting the 2008 housing crash, is shorting Nvidia, highlighting opaque accounting and circular funding patterns. He questions true end-user demand, noting most customers are supported by dealer financing, with OpenAI being central but lacking clear auditors. Even industry leaders acknowledge overexuberance. OpenAI’s Sam Altman admits investors are overly excited, yet also labels AI the most important development in a long time. Google CEO Sundar Pichai concedes “irrationality” exists in the current AI market. In sum, while AI’s transformative potential is widely recognized, massive capital inflows, heavy debt, complex financing structures, and circular deals raise valid concerns about sustainability and the risk of a looming financial correction.